The Rust Foundation's bad draft trademark policy

[diziet]

tl;dr

The Rust Foundation’s proposed new trademark policy is far too restrictive, and will cause (more) drama unless it is substantially revised.

• Process
• Substance
• Values
• Next steps
• Echoes of a dispute from 2006

Process

“Rust” is a trademark owned by the Foundation.

The Rust Foundation still seems to be finding its feet. Evidently, one of the items on its backlog was to update the trademark policy. Apparently they have been working on this for some time, in an informal working group.

In August, there was a survey. (I saw it in This Week In Rust, the community-curated newsletter where most important stuff appears, and responded.) I don’t think the results of this survey have been published anywhere.

Last week (12th April) the Foundation published an official Inside Rust blog post linking to a draft. They included a link to a feedback survey, closing on the 17th of April i.e., it was open for 5 days.

This is far too short a period for formal feedback on such a draft. Especially given that this process has apparently already been generating significant controversy within parts of the Rust community.

Substance

Overall, this policy is poor. It is far too restrictive.

It is likely to lead to (further) controversy and argument, including conflicts with Rust’s downstreams. It does not serve the needs of the Rust community.

In particular, the Rust community does not need the trademark to:

• “Maintain the quality” of software called Rust.

• Prevent people distributing badly-modified versions of its software, even called “Rust”, provided that users know what they are getting.

The community does need the trademark to:

• Prevent people impersonating the Rust Project (for example by registering competing domain names).

• Prevent people economically exploiting the value of the brand by holding shabby or vacuous conferences etc. (so requiring approval for e.g. conferences makes sense).

It might be useful to use the trademark to strengthen licensing or CoC compliance. For example, good faith redistributions of a modified Rust, as “Rust”, would be Free Software, even though the copyright licence permits proprietary derivatives; so use of the Rust trademark should probably require use of a Free licence.

There should be a series of blanket permissions to use the word “Rust” in for example:

• Free Software distributions derived from official Rust software, provided it’s recognisably similar. (I.e., you must change its name if you make other big changes too.)

• Free Software libraries written in Rust, or interfacing to Rust, in non-Rust-specific software repositories (including source hosting systems like gitlab, distro package managers e.g. Debian apt, and package managers for other languages e.g. npm).

Currently there aren’t. For example the current Debian practice of calling Rust libraries rust-<name-of-crate> is probably in violation.

There are a number of more detailed problems with the wording.

Values

The policy has all the hallmarks of excessive influence from traditional trademark lawyers and not enough influence from the Free Software community.

I would like to remind the Free Software activists on the inside of this process that the lawyers are there to serve you and the community.

The values embodied in trademark law often conflict with the values of the Free Software community. The Rust Project should adopt a trademark policy which follows the community’s values - even if that might weaken our ability to sue evildoers.

Next steps

The Foundation should take a step back and pause the process.

Then, the Foundation should restart the process from a much earlier stage, with much wider publicity. Each stage should be widely advertised to the whole community, with opportunities for feedback.

This should include publishing the results of the August 2022 survey. The Foundation should publish a sketch of the legal advice they have received, publicly say what the plausible options are and what their consequences might be (for the community, for downstreams, and for the Foundation’s enforcement ability).

(Some of this will no doubt repeat the work that has been done in the informal trademark working group. That work wasn’t widely enough advertised.)

Echoes of a dispute from 2006

Mozilla made a very similar mistake with Firefox in 2006.

The official policy stated that no-one was allowed to distribute Firefox with any patches, unless those patches had been pre-approved by Mozilla.

Debian is committed to Software Freedom. This must includes the freedom to modify the software as one sees fit, even if the original authors don’t agree.

Now, overly-restrictive trademark policies are hardly new. Debian often takes the practical view that usually the upstream with such a policy doesn’t really mean it.

But Mozilla decided they did mean it. They contacted Debian asking for Debian to get their patches approved.

Since that wasn’t acceptable to Debian, they stopped using the word “Firefox”. For a decade, Debian’s Firefox browser was called “Iceweasel”.

We don’t want something similar happening to “Rust”.